6 Powerline-Based Home Networks

In this chapter...

Introduction to Powerline Communications 88

Powerline Modem Technology 89

Technical Obstacles of In-Home Powerline Networks 91

The Criteria for Testing Powerline Solutions 96

Remote Management Features 98

In-Field Product Testing 99

Enikia Incorporated 100

X-10 107

Intellon CEBus 108

Inari Powerline Networking Technology 109

Summary 114

88 6 Powerline-Based Home Networks

Powerline-based home networking is an emerging technology that allows consumers to use their existing electrical wiring system to link appliances to each other and to the Internet. Home networks that use high-speed powerline technology can control anything that plugs into an outlet, including lights, televisions, thermostats, and alarms. The most popular powerline technologies are explained in this chapter.

INTRODUCTION TO POWERLINE COMMUNICATIONS ......

For several decades, researchers have attempted to use AC powerlines to create a communications network. Since almost all electronic devices already connect to AC powerlines for the electricity they need to operate, it seems only logical to develop a technology that would send data signals over the same wires. Considering that a vast majority of people in the world, even in developing areas, already live in homes with pervasive access to electrical outlets, such a technology would provide a quantum leap forward in the mass-market proliferation of IT and communications products. Using electrical wires seems logical and efficient. Traditional communications networks, such as phone lines, cable television, and computer data networks, use dedicated wiring designed specifically for communicating information. Powerline networks, on the other hand, were designed to deliver electricity, not data signals. This difference is not trivial. The highly variable and unpredictable levels of impedance, signal attenuation, and noise combine to create an extremely harsh environment that make high-speed data transmissions over powerlines very challenging. The following sections provide an overview of these technical obstacles and present some of the strategies that different developers are using to overcome them. But first, we will define the powerline network and present a background of how this industry has developed.

Industry Background

The powerline is not an ideal environment for data communications. Historically, it has proven much easier to modify the existing phone-line and cable networks for the modern needs of a digital economy. By upgrading portions of those networks with new digital communications equipment, the same copper phone lines and coaxial cable lines can be used for transmitting high-speed data trafﬁc. Although it is easier for companies to upgrade existing wiring, phone and cable wiring is not as widespread as electrical networks, especially outside of the United States. And while phone and cable networks might be effective at bringing Internet access to the home, they do not provide networks within the home. In the near future, having ubiquitous access points

Powerline Modem Technology 89

within the home will be increasingly important, especially with the proliferation of non-PC devices, or information appliances. And although wireless technology will be needed for battery-powered mobile devices, the majority of devices in the home re- main stationary and connected to the AC powerline network. It is for these reasons that high-speed powerline technology represents one of the most important hurdles in the world of communications and computing. But why weren’t powerline technologies available earlier? The next sections explore the powerline network and the inher- ent obstacles that the medium presents.

Deﬁning the Powerline Data Network

In conventional terms, the powerline connects the home to the electric utility company in order to supply power to the building. But powerline communications falls into two distinct categories: access and in-home. Access powerline technologies send data over the low-voltage electric networks that connect the home to the electric utility provider. The powerline access technologies enable a “last mile” local loop solution that provides individual homes with broadband connectivity to the Internet. In-home powerline technology communicates data exclusively within the consumer’s premises and extends to all of the electrical outlets within the home. The same electric outlets that provide power will also serve as access points for the network devices. Although the access and in-home solutions both send data signals over the powerlines, they are fun- damentally different technologies. Whereas the access technologies discussed in Chapter 3 focus on delivering a long-distance solution, competing with xDSL and broadband cable technologies, the in-home powerline technologies focus on delivering a short-distance high-bandwidth solution (10 Mbps) that would compete against other in-home LAN technologies, such as phone line and wireless. We will limit the discussion of powerline technology to the following deﬁnition of the in-home powerline network. The in-home powerline network, shown in Figure 6.1, consists of everything interconnecting through power outlets, including:

• House wiring inside of the building • Appliance wiring (power cords) • The appliances themselves (load devices) • The circuit breaker

POWERLINE MODEM TECHNOLOGY......

Communicating data over the powerline, just like in any other analog medium, requires some type of modulation device, or modem, that can transmit and receive data

90 6 Powerline-Based Home Networks

Medium Low voltage voltage

Electric Local distribution utility transmitter

Access network

In-home network

e AC powe m r li o ne -h s

In breakers Meter Circuit

Appliance (load device) Line of demarcation

Figure 6.1 In-home powerline network

signals. In order to turn a powerline electric network into a data communications network, a transceiver must be used to transmit the data from the device across the powerline medium. Thus the transceiver sends and receives digital data in analog form using the electrical outlets that it is connected to. Similar to other home network technologies, such as phone line and wireless, these transceiver nodes will take the form of microchips that will be embedded directly into next-generation computers and smart devices. But ﬁrst generation powerline network products will have to provide backward compatibility for devices that were not originally designed for powerline communications. Wall modules containing powerline transceivers will likely be the ﬁrst products on the market. These small devices could plug into an electrical outlet and replicate the sockets, similar to a power strip, but with an embedded powerline transceiver. The wall modules couple the device to the electrical network using a standardized commu-

Technical Obstacles of In-Home Powerline Networks 91

nications input interface, like a USB or Ethernet port. This way even traditional devices, like desktop PCs with Ethernet cards, will communicate with one another over the powerline.1 In such a conﬁguration, different network devices could share data, control one another, and access each other’s resources. But to facilitate these applications, the network’s communication throughput must be high enough to reduce noticeable time delays, especially when the network environment is noisy. The following section addresses some of the key technical obstacles that have made powerline communications so difﬁcult.

TECHNICAL OBSTACLES OF IN-HOME POWERLINE NETWORKS ......

Typical data and communications networks (like corporate LANs) use dedicated wiring to interconnect devices. But powerline networks, from their inception, were never intended for transmitting data. Instead the networks were optimized to efﬁciently distrib- ute power to all the electrical outlets throughout a building at frequencies typically between 50 to 60 Hz. Thus, the original designs of electrical networks never considered using the powerline medium for communicating data signals at other frequencies. For this reason, the powerline is a more difﬁcult communications medium than other types of isolated wiring like the Category 5 cabling used in Ethernet data networks. The physical topology of the electrical network, the physical properties of the electrical cabling, the appliances connected, and the behavioral characteristics of the electric current itself all combine to create technical obstacles.

Signal Attenuation, Impedance, and Appliance Loading

Attenuation describes how the signal strength decreases and loses energy as it transmits across a medium. In the powerline environment, the amount of attenuation that a signal experiences is primarily a function of the signal frequency and the distance it must travel on the wire. However, recent ﬁeld studies show that attenuation is also af- fected by other factors, including appliance loading and impedance discontinuities.

1. In this case, two computers in a home could be networked together connecting their RJ45 cables from the Ethernet card into two wall modules that would then stream the Ethernet data across the powerline. Later versions would most likely involve embedding the powerline transceiver directly onto the motherboard of the PC, effectively eliminating the need for an Ethernet card and Cat5 cabling altogether. Instead the data would stream over the same cord that the computer uses to draw power.

92 6 Powerline-Based Home Networks

Distance between Socket Links • As a signal travels across a wire, it inevitably loses energy. Attenuation increases as a function of the distance the signal must travel on the wire. In other words, the signal gets weaker the farther it must travel between the transmitting and receiving devices. Frequency • Generally, the higher the frequency of the carrier, the greater the attenuation. Impedance Discontinuities • Impedance refers to the resistance in ﬂow of the alternating current (AC) in an electrical circuit. Impedance discontinuities, caused by wire nuts, switches, wall socket outlets and their appliance loads, create nulls, or spectrum “suckouts.” Wall sockets create a problem both as unterminated network points (when there is no device connected) and as appliance loading points (when there is a device connected). Network Load • The network load is determined by the number and type of appliances that are connected to the electrical network in relation to the network’s overall size. All of the electric devices in a consumer’s home, like televisions, lamps, washing machines, and other appliances, combine to change the impedance of the network at various points. Even devices that are not operating and consuming electricity still inhibit the network’s performance because they load the line both resistively and reactively, causing impedance mismatches that dissipate the energy of the signal as it travels across the network. New impedance mismatches are created every time a device is plugged in or out of the network. All of these factors combine to create the network load, which causes increased signal attenuation across the overall topology of the network.

Real-World Attenuation Scenarios

Overcoming signal attenuation is important for developing marketable products since a consumer may want to network two devices located at opposite sides of their home. If the signal attenuates excessively these devices may not be able to communicate at all, rendering the technology ineffective. But in order to successfully address the problem of attenuation it is necessary to ﬁrst understand some of the interrelated variables that affect how signals attenuate in real-world environments.

Attenuation as a Function of Network Loading and Home Size

Since attenuation increases with distance, it seems logical to assume that larger homes, with greater average distances between socket links, would present greater attenuation problems. But ﬁeld research shows that this is not necessarily the case. This is because the powerline networks tend to be less “loaded” in larger homes due to the

Technical Obstacles of In-Home Powerline Networks 93

fact that in larger homes there are more circuits with a similar number of household appliances connected. For example, imagine a family moving into a home twice the size of their previous home. It is unlikely that they will furnish their home with twice as many appliances as their previous home (they probably wouldn’t purchase an additional refrigerator and washing machine). Thus, given the same quantity of appliances, the overall network load per circuit in a larger home may actually be lower than that of a smaller home. As a result, the increase in attenuation caused by distance in the larger home may in fact be offset by the lower levels of each circuit’s load.

Attenuation as a Function of Network Loading and Transmission Frequency

Another consideration is that not all signals attenuate equally. For example, a high frequency signal will attenuate more rapidly than a low frequency signal that is transmitted over the same length of wire. The first graph in Figure 6.2 shows that the wire attenuation of a signal (caused by distance) will be greater if the signal is transmitted at higher frequency levels. As previously stated, appliance loading is a primary cause of signal attenuation. But the amount of attenuation caused by a given appliance load will also vary with re- spect to the transmission frequency of the signal. The second graph in Figure 6.2 dem- onstrates that the attenuation from appliance loading actually decreases at higher frequencies. This is due to the fact that most appliances have capacitive filters that significantly limit signals in lower frequencies. Therefore, as signal frequency increases above the limits of the capacitive filters in appliances, attenuation caused by appliance loading decreases. The final graph in Figure 6.2 shows a real-world attenuation environment where increased wire attenuation in higher frequencies is significantly offset by appliance load attenuation in higher frequencies.

Attn Attn Attn

+ =

f f 450+ MHz f Wire Appliance Real-world attenuation loading attenuation environment

Figure 6.2 Attenuation effects

94 6 Powerline-Based Home Networks

Interference and Noise Sources

In powerline communications, the term “noise” refers to any undesired signals on the wire that interfere with the data transmission of the original communications signal. The two common types of noise that affect the powerline environment are injected noise, caused by operating appliances, and steady state background noise.

Injected noise

Injected noise (also called impulse noise) results from the switching of inductive loads that produce strong impulses. The worst offenders producing injected noise are devices containing AC motors, such hair dryers, vacuum cleaners, and electric drills. When these appliances are turned on, they saturate the powerline with violent spikes of noise that defeat many communication methods (see Figure 6.3). For this reason, powerline technologies must focus on overcoming injected noise impairments.

Blender off Blender on Blender off

Amplitude

Time

Figure 6.3 Injected noise

Background noise

Background noise (also called ambient or white noise) presents an additional im- pediment. This quasi-steady state of noise is caused by radio frequency interference (RFI) from noise emitters such as ﬂorescent lights that may be operating in the vi- cinity of the network.

Technical Obstacles of In-Home Powerline Networks 95

Noise as a Function of Frequency

As with attenuation, the characteristics of noise interference change across different transmission frequencies. But while the attenuation problem increases at higher frequencies, the problem of noise interference decreases (see Figure 6.4). Thus the noise and attenuation problems present inevitable tradeoffs in the selection of an optimal transmission frequency band. Solving this “frequency dilemma” lies at the core of most strategies for developing a high-speed powerline solution. Noise amplitude

Frequency Noise energy density

Figure 6.4 Noise as a function of frequency

The Multireﬂective Effect (Standing Waves)

When a beam of sunlight shines at the surface of water, some of the light will go through the water, and some of the light will reflect off the surface, bouncing in different directions to create the shimmering effect that people see. The fact that the air and water are physically different media causes part of the light beam to reflect back toward the source of the signal. In powerline networks a similar phenomenon happens to the data signal, known as the multireflective effect. As stated previously, impedance is a measure of the resistance that the signal experiences as it travels across the wire. Just as when some of the light reflects as it passes from air to the water, some of the data signal reflects as it experiences impedance discontinuities on the wire. The multireflective effect comes into play anytime the signal changes impedance as it crosses a medium. Unterminated network points, changes in the physical wiring

96 6 Powerline-Based Home Networks

structure, and jumping phases at the circuit represent common impedance discontinuities in electrical networks. The multireflective effect presents an obstacle for communications because the portion of the signal that is reflected back along the origination path can interfere with the source signal. In these cases the reflected signal creates a standing wave that may result in a null, or cancellation, of the source data signal.2 Fortunately, research has shown that the nulls, which appear as spectrum suckouts across certain frequencies, occur as a function of the network’s topology. For example, regulations in electrical coding, like the distance between wall sockets, are consistent across the network. This makes it possible to exploit regularities and statis- tical characteristics to overcome certain aspects of the multireflective effect.

Demographic Variability of Electrical Networks

Just as homes vary greatly in their size and physical layout, the electrical networks that serve them also vary. The average and maximum distance between wall sockets (and subsequently node devices), the placement of the breaker panel and its effect on phase jumping, and the era of the home’s electrical system vary greatly across the demographics of the American and international marketplace. Consider, for example, two homes of equal size (3000 sq. ft.); one is single story and the other is two stories. The average physical distance between outlets, and subsequently the network devices, would be much greater in the single story home, making it more susceptible to the effects of attenuation and the multipath effect. Also consider the difference between a WWII-era American home, that might have an electrical network made from alumi- num wiring, and a modern home that uses copper wiring. And, as previously stated, the powerline home network inevitably extends to the power transformer. Thus, the number of homes connected to each transformer presents another consideration that varies greatly across residential demographics.

THE CRITERIA FOR TESTING POWERLINE SOLUTIONS ......

Evaluating the overall effectiveness of a given powerline solution can be difﬁcult, considering the multitude of technical obstacles and different possible approaches for overcoming each of them. The following three criteria are typically used to benchmark the technical performance of a given powerline communications solution. To-

2. The multireﬂective effect illustrates one of the key advantages that networks using dedicated wiring possess. Isolated cabling, like Category 5 for Ethernet, provides consistent impedance across the length of the wire, making the multireﬂective effect irrelevant.

The Criteria for Testing Powerline Solutions 97

gether, the bit error rates (BER), the throughput levels, and the latency levels of a given solution provide a solid benchmark for its effectiveness.

Technical Criteria

Bit Error Rate (BER)

The BER quantiﬁes the amount of data lost between the transmitter and receiver when the signal is propagated across the line. BER measurements are particularly important in powerline networks where the harsh noise environment can cause considerable loss of data packets. The BER ﬁgures provides a useful measure of the performance of different coding techniques and serve as a good initial indicator of data throughput.3

Throughput Levels

Nominal and effective bit rates provide an indication of a technology’s overall performance in normal to noisy environments. Because the powerline is inherently so noisy, it is inevitable that the throughput varies with the characteristics of the environment. For example, when a blender (or other brush motor device) is being used it will inject signiﬁcant noise onto the line, degrading the effective throughput of data. Some technologies might be completely unable to transmit data in such an environment, effectively crashing the network, while others might be able to transmit data consistently, but at much lower throughput levels.4 This issue is considerable if you extend the blender example to all of the other devices in a home, like hair dryers, heaters, and stereos, that consumers will use without concern for their effect on the performance of the network. For this reason, both nominal and effective bit rate measurements together provide a true assessment of a technology’s overall functional performance under both the ideal and noisy environments of the real world.

3. Although the effective and nominal throughput levels are typically a function of the Bit Error Rate levels, the level of correlation between them can vary with different technologies. For this reason they are identified as separate criteria, although they tend to be highly interdependent. 4. One of the key requirements for a home network is “dial tone” quality, whereby the network never crashes. Instead the network performance will “degrade gracefully” and bandwidth will scale down in the face of noise offenders. Thus, a high-speed network with a nominal throughput of 10 Mbps must be able to scale down to a lower effective throughput (perhaps to a minimum of 1 Mbps), which could still manage high-priority traffic even with significant environment noise. Quality of Service (QoS) features will be necessary for high-speed networks as a way of priori- tizing data between devices that compete for bandwidth. For example, QoS parameters could help ensure that, in a noisy environment, the packets from a voice conversation by one user have priority over the packets of Web page downloads of another user on the same network.

98 6 Powerline-Based Home Networks

Latency Levels

This metric is used to assess the time lag for data to travel between nodes (transmitter and receiver) on a network. Although the latency requirements for a home network vary depending on the applications needs, a latency level of about 10 milliseconds is necessary to support synchronous (real-time) applications, like voice telephony or video conferencing. Anything greater than this level would cause a noticeable delay that many consumers would likely ﬁnd unacceptable.

Functional Criteria

While the previous criteria can help determine how effectively a product performs technically, the following attributes may be equally important for vendors and OEMs who need to formulate implementation strategies to deploy the technology into the actual consumer marketplace.

Quality of Service (QoS) Features

In the powerline environment, it is nearly impossible to maintain the throughput with noise levels that fluctuate over time. Even high-speed technologies that might attain 20 Mbps nominal throughput will be significantly reduced during periods when multiple appliances are injecting noise onto the line simultaneously. Even though such oc- currences might be rare, a consumer would find it unacceptable if the network crashes. And while they may tolerate slower Web page downloads for a short period, they will not tolerate having their phone conversation break apart. For this reason, it is important that the technology allow quality of service levels to be independently as- signed to each device. In this way, the data packets of a voice call would have priority over the data packets from a print job. During times when there is limited bandwidth, the QoS scheme will help sustain the higher priority applications, making the network more reliable and effective from the user’s viewpoint.

REMOTE MANAGEMENT FEATURES ......

Many technology proponents believe that home networks will only penetrate the mainstream marketplace once they are deployed by service providers who use them to extend their broadband Internet connection into the home. It is entirely possible that service providers may eventually choose to redeﬁne the lines of network demarcation to actually include the home network itself. Yet for this to happen, the service provider would need the ability to query a network that now extends all the way to the power-

In-Field Product Testing 99

line transceivers embedded either into wall modules or into the devices themselves. Remote management features would need to be integrated into the transceivers in order to give network operators and service providers the option of this functionality. One of the largest single expenses that service providers incur is the “truck roll” of dispatching a technician to the user’s home for maintenance. With remote management features, intelligent “wizards” could be used to automatically conﬁgure the network and test it at periodic intervals. It would also enable the service provider to remotely query the network from a centralized customer service center. Thus, the more remote management features a powerline technology supports, the more likely it is to be adopted by service providers who are deploying broadband solutions.

IN-FIELD PRODUCT TESTING ......

In-field testing represents an important step in the development of any communications product because it helps to prove the “real-world” viability of the technology. But it is particularly important for powerline technologies, which are specifically designed to overcome the randomness and complexity of the environment. In essence, the challenge to powerline communications is to design a product that works in an environment that was never intended for such use. In other words, sending a data signal down a copper wire that carries an electrical current is one hur- dle. But it is quite a different technological challenge to produce a universal, low-cost powerline transceiver that operates reliably between all sockets links, in all homes, at all times throughout the day. The following list highlights some of the key obstacles that a powerline technology will encounter during field trials.

Socket Links

The transceivers must be able to establish connections and maintain transmission bandwidth across all combinations of socket links in a given home. This means that signals may have to jump phases at the circuit breaker and reach opposite sides of the home without attenuating signiﬁcantly.

Time Variability (Appliance Usage Cycles)

The transceivers must be able to maintain transmission bandwidth consistently over the time cycles that correspond with appliance usage. This means that bandwidth transmission must be maintained even as appliance usage varies throughout the day. 100 6 Powerline-Based Home Networks

Such usage patterns may involve multiple sources of noise being injected onto the network simultaneously. (Imagine the powerline environment in a home with different family members operating a dishwasher, a blender, and a hair dryer at the same time, compared to that same home when everyone is at work or school.)

Household Demographics

The transceivers must operate across the vast majority of homes in the marketplace. This involves signiﬁcant variability in size, design, and materials used for the electrical network. Although in the past many developers were able to achieve high-speed data communications in laboratory environments, none of these technologies evolved into consumer products because they were unable to successfully pass through the stage of in-ﬁeld testing. The following sections cover some of the newest advancements and innovations from developers in the powerline communications industry, which may bring products enabling speeds of 10 Mbps and beyond as soon as early 2001.

ENIKIA INCORPORATED ......

Enikia Incorporated is a privately held company headquartered in Piscataway, New Jersey. A start-up founded in 1997, Enikia quickly established itself as a technological innovator and industry leader by being the ﬁrst company to publicly demonstrate that high-speed in-home powerline communications was possible. After announcing that its upcoming product line would enable speeds of up to 10 Mbps, Enikia followed up by successfully unveiling its prototype units. Enikia’s product line consists of powerline Ethernet transceiver chipsets that enable data transmission speeds of 10 Mbps and above. Device OEMs purchase Enikia’s chipsets or license Enikia’s intellectual property in order to embed this technology into intelligent devices. Enikia’s solution makes it possible for computers, as well as other “smart” appliances, to communicate with one another over the home’s electrical network.

Enikia’s Research and Development Efforts

Enikia’s technology team invested two years of R&D to realize their vision of an ele- gant home networking solution that was reliable, inexpensive, and easy for the end user to operate. During the R&D period, Enikia’s team conducted in-ﬁeld studies of the powerline environment, gathering research data from real-world homes through- Enikia Incorporated 101

out the United States and in international locations. These studies provided some of the following discoveries that led to Enikia’s novel approach for developing a high- speed powerline solution:

• The powerline is capable of reliably transmitting a signal over a wide range of frequencies. • Noise impairments on the line can be defeated by adapting data transmission techniques to the changing environment over time. • A combination of wiring methods can be used to transmit high-speed data signals.

By rigorously studying the effects of noise generators and appliance loading in actual homes, the Enikia team was able to statistically model these real-world environments. This knowledge was then integrated into a protocol, called ACT, which represents the core of Enikia’s solution for powerline communications.

The Noise Generator and Appliance Loading Studies

The noise generator studies characterized the properties of injected noise emitted by different generators (such as a hair dryer versus a light dimmer). These studies showed that the continuous noise injected by many appliances creates a strobing effect that equates to periodic noise spikes in the time domain. (See Figure 6.5.) Enikia was able to statistically model the pulse widths and separation of the spikes and found that the time period between spikes offers an opportunity to transmit data across the line. To exploit this opportunity, Enikia’s protocol design takes a traditional Ethernet frame, segments it into small packets, and sends them in the clear time spaces between the impulse noise spikes. On the receiving side, Enikia’s protocol then reassembles the segments into the original Ethernet frame, making the segmentation and reassembly process transparent to devices connected to the network. Enikia also discovered that a combination of different wiring methods could be used. Most powerline technologies use the hot-neutral wires only. But Enikia’s studies showed that the hot-ground wires could also be used for transmitting signals, and this in fact often presented a very clean channel for communication. In the past, higher frequencies were not considered viable for powerline communications, due to the high attenuation levels. But, Enikia’s ﬁeld studies showed that frequencies in excess of 100 MHz can be reliably used for homes up to 5000 sq. ft. In addition, Enikia’s technology includes a repeater function that extends the reach of the network. 102 6 Powerline-Based Home Networks

Blender off Blender on Blender off Amplitude

Time

Figure 6.5 Effect of continuous injected noise

In the United States, the FCC imposes limits on the amount of energy that can be injected on the powerline at different frequencies. The main purpose of this is to protect the rights of licensed spectrum users from interference from unintentional ra- diation. Enikia’s signal was tested within the FCC-imposed energy limits and showed that a signal could exist on the powerline in a wide variety of frequencies without in- terfering with licensed users. Enikia Incorporated 103

Development of Enikia Core Technology

The beginning of this chapter identified the powerline environment as an extremely difficult communications environment. Background noise from outside interferers, injected noise from in-home appliances, multireflective effects, and widespread signal attenuation are the primary impairments to communications. Enikia engineered a powerline technology, embodied in the following components, that addresses each obstacle and enables high-speed communications even in the harshest environments.

• The protocol processor—The protocol processor runs Enikia’s ACT protocol and SST token-passing scheme. • The DSP modem—Enikia’s DSP modem is comprised of 16 digital channels and employs DBPSK and DQPSK selectively according to channel conditions. • The analog RF front end—Enikia employs a low-cost simple analog design to couple the protocol processor and DSP modem to the powerline.

Figure 6.6 illustrates, and the following sections detail, these major elements of Enikia’s technology.

ACT Protocol Overview

ACT stands for Adaptive to Channel and Time. The protocol consists of a frequency dimension, which evaluates the environment and determines the frequencies that provide the clearest path for the signal. It also includes a time dimension, which allows the signal to adapt to the ever-changing conditions on the powerline. When viewing impulse noise on an oscilloscope, it appears as lightning strikes, where there is a clear space between every impulse. Since the noise is narrow, the network is clear for a time period between the noise spikes. In the same way that WWI planes used to shoot their machine guns time-synchronized with the rotation of their propeller blades, Enikia is able to send short packets through the clear spaces in between the impulse noise strikes. These short bursts of data are delivered very quickly. The ACT protocol uses an environmental detector called the Received Signal Strength Indicator circuit (RSSI). This circuit aids the transmitter in identifying the beginning of a “clear” time interval as the preferred instance to launch the data across the wire. Thus, the data has a better chance of avoiding noise and maintain- ing a reliable link. The ACT protocol also uses techniques to sense the density of noise on the line and adjusts the packet size and error-correction schemes accordingly. Enikia calls 104 6 Powerline-Based Home Networks

Figure 6.6 Enikia’s technology components

these different packet sizes gears since they are representative of shifting gears in a car depending on the conditions of the road. By varying the packet size in each gear, the communication throughput is optimized. Under extreme noise conditions, the ACT protocol employs lower gears containing extensive error correction and small packets to fit in between the noise spikes. As the noise environment becomes less volatile, the ACT protocol shifts to higher, more efficient gears with larger packets sizes and less error correction. The technology also uses up to 16 channels in parallel over a 20 MHz range. The channels adapt to the changing environment, and only the channels that communicate clearly are used at any given instant. For example, all 16 channels may be communicating at once, delivering a full 20 Mbps between transceivers. But when noise is present, a spike may flow through the 20 MHz range and eliminate the use of a few channels. The ACT protocol adapts the flow of data by eliminating certain channels, changing to a shorter packet length, changing its modulation technique, and slowing the speed of transmission so that a reliable link is maintained. Enikia employs an additional technique, known as HOP (Historically Oriented Preference), which allows two transceivers with a poor communication link to reroute Enikia Incorporated 105

their signal through a third transceiver. HOP relates to how XCVR A and XCVR B might use another XCVR as a kind of repeater if taking that path offers better communications. HOP uses historical data to make decisions about which path through the network is best. Since powerline networks can vary greatly moment-by-moment or in a certain daily or weekly cycle, keeping this information is valuable. By frequently evaluating the environment on the powerline, and establishing a mode of communication based on the state of the communications medium, Enikia’s ACT protocol succeeds in making the network reliable.

Secure Sparse Token (SST)

The protocol processor also runs a Medium Access Protocol called Enikia’s Secure Sparse Token (SST) token-passing scheme for Quality of Service considerations. The SST scheme was speciﬁcally designed for the powerline environment due to its quick token regeneration, special priority of token passing, and robustness for real-time data priorities. Enikia’s SST algorithm is based on the IEEE 802.4 token bus protocol, which has been modiﬁed by Enikia for the powerline environment. In general, token bus offers some unique properties that are a good match for the home networking environment:

• Short token-passing times for home networks (characterized by a small diameter) • Different levels of priorities for data • Deterministic delay properties • Contention-free environment • Reduction in peak signal levels due to lack of contention

In addition, the SST provides several features that are especially desirable and often critical for the home network powerline environment. For example, the SST supports local, small-size (7-bit) addresses, as economical fronts for full-size (48-bit) Ethernet addresses. However, all nodes are aware of both (local and Ethernet) addresses of the other nodes in the network. For networks that use Ethernet hubbing, SST supports full addressing of the nodes that may reside in a network section behind a hub. Further, to increase the efﬁciency of communicating nodes on the network, SST allow inactive nodes to drop out of the token passing scheme. This allows only the currently active and communicating nodes to “join” in a conversation, decreasing latency and increasing available bandwidth. 106 6 Powerline-Based Home Networks

In addition, SST allows the active ring to become totally silent, eliminating token passing entirely, when no node is active. Security is addressed by using very secure three-way handshaking for token passing, and efﬁciency is addressed by using small control frames (approximately 30 milliseconds).

Digital DSP Modem

Enikia’s DSP was designed to accomplish the following objectives:

• Be able to ﬁt symbols in between noise spikes by utilizing fast symbol rates • Defeat “nulls” by using narrow channel widths • Utilize a minimum number of channels to achieve low production cost

Taking these criteria into account, Enikia engineered its DSP with channels using swift symbol rates to take advantage of the impulse noise time domain spaces, narrow channels to accept and conquer extremely deep nulls, and a limited number of channels to keep the DSP inexpensive for consumer applications. Further, the design overcomes narrowband interferers. Most narrowband interference will only affect a single channel, rather than affecting multiple carriers or the entire signal. The DSP Modem also implements two modulation schemes: DBPSK and DQPSK. These are used selectively according to the varying powerline environments. Under the harshest SNR (signal-to-noise ratio), the modem uses DBPSK because of its high resiliency to noise. But when signal-to-noise ratios improve, the modulation scheme switches to DQPSK for increased efficiency, doubling the potential throughput. Therefore Enikia’s technology is able to balance the tradeoffs between throughput and reliability as channel conditions fluctuate. Future versions of Enikia’s technology are able to employ even more efficient versions of quadrature amplitude modulation (QAM), allowing more bits per Hz and greater throughput.

Analog RF Front End

Enikia uses a low-cost, simple analog design for its RF front end. Since the DSP is responsible for the channel signal processing, Enikia only had to include a single A/D, D/A in its design. The efﬁciencies of the ACT protocol allow Enikia to inject low power onto the line, and therefore, the RF front end is able to use these three bands: low band (2 to 30 MHz), mid band (108 to 174 MHz), and high band (216 to 470 X-10 107

MHz). Even though the mid and high bands are more efﬁcient radiators, the low injected power keeps Enikia’s signal well within FCC guidelines for such frequencies. Because of the spectrum available in these three bands, future Enikia products will offer very high throughput communications, allowing multichannel video distribution, and other very high bandwidth applications. Further, since the RF front end’s design is so simple and uses relatively few components, it allows easy design of an analog semiconductor.

X-10 ......

X-10 is a communications protocol that allows compatible home networking products to talk to each other via the existing electrical wiring in the home. Basic X-10 powerline technology is almost 20 years old and was initially developed to integrate with low-cost lighting and appliance control devices. X-10 originally started out as unidirectional only; however, capability has recently been added for bidirectional communication if needed. Nevertheless, the vast majority of X-10 communication remains unidirectional. X-10 controllers send signals over existing AC wiring to receiver modules. The X-10 modules are adapters that connect to outlets and control simple devices. X-10 powerline technology transmits binary data using an amplitude modulation (AM) technique. To differentiate the symbols, the carrier uses the zero-voltage crossing point of the 60 Hz AC sine wave on the cycle’s positive or negative transition. The zero-crossing point usually has the least noise and interference from other devices on the powerline. Synchronized receivers accept the carrier at each zero-crossing point. To reduce errors, X-10 requires two zero crossings to transmit either a zero or a one. Therefore, every bit requires a full 60 hertz cycle and thus the X-10 transmission rate is limited to only 60 bps. A complete X-10 command consists of two packets with a 3- cycle gap between each packet. Each packet contains two identical messages of 11 bits (or 11 cycles) each. Therefore, a complete X-10 command consumes 47 cycles that yields a transmission time of about .8 seconds. Using X-10 it is possible to control lights and virtually any other electrical device from anywhere in the house with no additional wiring. The X-10 technology and resource forum designs, develops, manu- factures, and markets products that are based on this standard. Today, scores of manu- facturers make X-10-compatible products that, at $10 to $30, scarcely cost more than their incompatible counterparts; according to the X-10 group, more than 100 million such products have been sold. These home automation products are called “powerline carrier” (PLC) devices and are often installed by builders who want to offer home automation as an additional selling feature. The home automation line consists of “controllers” that automatically send signals over existing electric power wiring to receiver “modules,” which in turn control lights, appliances, heating and air conditioning units, 108 6 Powerline-Based Home Networks

etc. With the X-10 standard, you can literally walk into a nearby electronics store and purchase all of the necessary equipment to automate your home with the X-10 standard. The main disadvantage for legacy X-10 technology is that it has very limited capability in terms of both speed and intelligence. It is a technology relegated to control applications only because of its low data rate and rudimentary functionality.

INTELLON CEBUS ......

Intellon Corporation is a leading supplier of communication solutions for networking with no new wires. Intellon provides low-cost, high-performance digital, mixed-signal, and radio frequency (RF) based systems on silicon to OEMs worldwide. Intel- lon’s intellectual property, products, technologies, and services are key to a variety of open industry standards and fast-growing markets including telecommunications, networking, transportation, and consumer electronics. Intellon’s powerline carrier and radio frequency technologies enable high-speed communication and extend the reach of the Internet to individual products without adding new wires. Founded in 1989, Intellon is a privately held company based in Ocala, Florida, and operates as a fabless semiconductor company to develop, manufacture, and dis- tribute integrated circuit-based products and modules that are supported by complete reference design information, development, and evaluation tools. The company pro- duces products that conform to the Consumers Electronics Bus (CEBus) standard. The CEBus standard is an open standard that provides separate physical layer speciﬁ- cation documents for communication on powerlines and other media. The Intellon technology is oriented toward providing control capabilities to home networks and consists of two fundamental components—a transceiver implementing spread spectrum technology, and a microcontroller to run the protocol. (The next chapter explains spread spectrum technologies in greater detail.) Data packets are transmitted by the transceiver at about 10 Kbps, employing spread spectrum technology. Each packet contains the necessary sender and receiver addresses. The CEBus protocol uses a peer-to-peer communications model so that any node on the network has access to the media at any time. To avoid data collisions, it uses a Carrier Sense Multiple Access/ Collision Detection and Resolution (CSMA/CDCR) protocol. Basically, this Media Access Control (MAC) protocol requires a network node to wait until the line is clear, which means that no other packet is being transmitted before it can send a packet. A CEBus network is comprised of a control channel and potentially multiple data channels on each of the CEBus media. CEBus control channel communication is standardized across all media, with a consistent packet format and signaling rate, and is used exclusively to control devices and resources of the network, including data channel al- locations. A control channel signaling rate of 10,000 unit symbols per second is used with a “one” bit taking one unit symbol time (UST) and a “zero” bit taking two UST, Inari Powerline Networking Technology 109

resulting in an effective control data rate of approximately 7,000 data bits per second. Data channels typically provide selectable bandwidths that can support high data rates and are used to send data such as audio, video, or computer files over the network. The characteristics of a data channel can vary greatly depending on the medium and connected device requirements. All data channel assignments and functions are managed by CEBus control messages sent via the control channel. CEBus includes a common application language (CAL) that allows devices to communicate commands and status requests between each other using a common command syntax and vocabulary. CAL defines various electronic device functional subunits called contexts. For example, the audio control of a TV, a stereo, a CD Player, or a VCR is a CAL context. Each context is further broken down into objects, which represent various control functions of the context; for example, volume, bass, treble, or mute functions. Finally, objects are defined by a set of instance variables that specify the operation of the function of the object, such as the default or current setting of the volume object. Further details of CAL and CEBus are available in Chapter 15. By using the CAL specification, Intellon ensures their chips can communicate with other CAL-compliant devices. Intellon offers products ranging from chip sets to board solutions, depending on the level of integration the manufacturer wants to perform on their own. In addition to a wide variety of powerline-based products, Intellon also offers customers a range of RF based products. Just before the finalization of this guide, the HomePlug Powerline Alliance announced the selection of Intellon’s Tech- nology as the basis for its industry specification for powerline home networking. HomePlug Powerline Alliance, Inc. is a nonprofit organization established to provide a forum for the creation of specifications for home powerline networking products and services, and to accelerate the demand for these products and services through the sponsorship of market and user education programs. Founding HomePlug member companies include: 3Com, AMD, Cisco Systems, Compaq, Conexant, Enikia, Intel, Intellon, Motorola, Panasonic, S3’s Diamond Multimedia, Tandy/RadioShack and Texas Instruments. For more information, see Intellon’s Web site at http://www.intellon.com.

INARI POWERLINE NETWORKING TECHNOLOGY ......

Inari was founded in 1997 and is headquartered in Draper, Utah. Inari (formerly Intel- ogis) is a leading developer of powerline networking technologies. The company’s core technology, which transmits high-speed digital signals over existing AC electrical wiring, was developed by its engineers while the company was part of Novell’s Embedded Systems Technology (NEST) division. As a product company, Inari be- came well known for its PassPort Plug-in Network home networking product. Today 110 6 Powerline-Based Home Networks

Inari is focussing its efforts on providing powerline chipsets to modem, gateway, and network interface card vendors as well as to OEM partners in the consumer electronics marketplace. The next section of this guide will describe the technical architecture of Inari’s powerline networking technologies.

Inari Technical Architecture

Today, Internet access is a paramount concern for customers in the home and home- ofﬁce market segment. Over 76.5 million homes in the United States have an Internet connection and that number is growing exponentially. As mentioned earlier, an even more interesting fact is that approximately 20 million homes have two or more computers competing for time on the Internet. These factors have driven the current demand for the ability to simultaneously share a single Internet connection within the home. Close on the heels of the Internet fascination will be the convergence of home entertainment and the home network. Technologies such as browsing the Internet from your TV or downloading your favorite songs off the Web are intriguing, yet they have not managed to fully capture the attention or appetite of the home consumer. While the computing industry might have not yet hit on the right product offering, most technology pundits agree that one of the next great crazes will deal with distribution of streaming video and audio data in the home. Experts also agree that once the challenges of streaming data have been overcome and become widely accepted by consumers in the home that the door will open for the home automation market. For home automation to succeed, networking technology needs to not only be inexpensive, but it has to reach a level of simplicity that non-computer users can implement and use. Streaming data solutions that enable voice activated control of home appliances will go a long way to achieve the necessary levels of simplicity. All of the above factors point to the conclusion that decreasing networking technology costs and avail- ability of requisite networking technologies will drive the home and small business network market segment through three general stages of evolution:

• Distribution of Internet data • Distribution of streaming data (audio and voice) • Home automation

To meet the new requirements of these evolving marketplaces, Inari has created a set of networking technology components that allow solution providers to build products that take advantage of the ubiquitous powerline network medium at currents speeds up to 2 Mbps and future speeds in excess of 10 Mbps. The following sections detail the major portions of Inari’s powerline networking technology (PNT). Inari Powerline Networking Technology 111

Digital Powerline—Physical Layer

As mentioned, two of the main drawbacks of most powerline technologies are high electronic component expense and low data throughput rates. Inari created the Digital Powerline (DPL) protocol to overcome these obstacles that have historically stood in the way of communicating via powerlines. Anyone who has a general understanding of the powerline medium and its underlying physical layer realizes that it exhibits a very dynamic behavior. As loads are added or removed from the system, or as devices power on or off, the system characteristics change. Two of the more important characteristics that the physical layer and Media Access Control (MAC) layer must be aware of are time-varying attenuation and noise. Inari’s Digital Powerline was created to address these issues at the physical layer. DPL delivers a set of rules that deﬁne how information is transmitted between the physical components of the network. The speciﬁc physical components of DPL are transceivers and the powerline cabling itself. The circuitry required for Inari DPL modulation is scalable, meaning that the number of data channels implemented is determined by the applications’ need for throughput, which projects directly into affordability. Far fewer and less expensive components are required to implement Inari’s modulation scheme than those needed for spread spectrum modulation. Furthermore, Inari’s DPL delivers low bit-error rates in the range of 10-9, thus making DPL well suited to provide robust high-speed network service to any home and to many commercial environments as well.

Powerline Exchange Protocol—MAC Layer

Many devices residing in a home or small ofﬁce network require not only deterministic, periodic access with guaranteed time slots, but also minimal latency between token rotations. One of the most powerful features of Inari’s Powerline Exchange (PLX) protocol is its ability to provide deterministic time slots for streaming applications and devices. PLX can be characterized as a hybrid protocol that utilizes Datagram Sens- ing Multiple Access with a Centralized Token-Passing scheme, or DSMA/CTP. PLX was built from the ground up to support control systems and data networking devices as well as deterministic audio and video streaming applications. It features numerous quality of service (QoS) hooks that can be exploited by application developers, including the following: • Prioritization by device type • Packet life expectancy variability • Flexibility in bandwidth allocation (multiple time slots) • Intelligent channel management capabilities 112 6 Powerline-Based Home Networks

• Reuse of unused time slots • Real-time channel status information • System throttling (packet-pacing) mechanism • Minimized worst-case latencies for highest priority devices • Packet bursting within a token session PLX also incorporates a robust encryption algorithm as well as a mechanism to seed remote devices with the 32-byte encryption array (Difﬁe-Hellman handshake). The following outlines some of the other key features of PLX: • Simple datagram (preamble) detection with random back-off for multin- ode contention phases • Active server for prioritized, centralized token-passing phase • System throttling (packet pacing), smart packet/channel retries • QoS hooks, including multichannel registration • 32-bit static node addressing, with 16-bit cyclic redundancy checking • Ad hoc networking, multirate support, and security • Guaranteed 64 Kbps timeslots for isochronous communications • Small packets to facilitate low-latency switched circuits • Normally “quiet” medium, only active nodes communicate • Constant aggregate throughput characteristics • Thin network and transport layers to support embedded applications Since multiple benched players may contend for the same time slot, a random back-off mechanism is employed, to eventually isolate a single node. This contention period is similar to Ethernet’s random back-off period.

Inari Common Application Language—Application Layer At the application layer Inari provides a complete client-server solution. On the client side, Inari has adapted a hybrid of CEBus’ Common Application Layer (CAL) that can reside on the network nodes. Still, the PNT architecture allows OEMs to implement other client solutions to interface with their network node’s application. Inari also provides a rich set of server-side technologies that OEMs can take advantage of as needed. Inari Implementation

As mentioned, Inari PNT has adapted the CEBus Generic Common Application Language as one of its application layer protocols for networked nodes with one major exception: CEBus Generic CAL uses a peer-to-peer topology, and Inari CAL Inari Powerline Networking Technology 113

utilizes a client-server topology. Using a client-server topology allows more of the intelligence of each Inari node’s application to be placed in a centralized application server, such as an existing computer or similar device, instead of residing within each individual client node. On the other hand, the peer-to-peer environment of CE- Bus’ generic CAL requires each node, however simple it might be, to contain information regarding all other network nodes. In a peer-to-peer environment, each node must contain the requisite circuitry, processing power, memory, protocols, and software intelligence to perform much of the application tasks that are performed by a single application server node in the Inari client-server architecture. Furthermore, in a peer-to-peer environment, each node must also be aware of all remote nodes with which it needs to communicate, as well as know about the capabilities and attributes of each of those nodes. With the Inari architecture, each client node is totally unaware of the existence and capabilities of other nodes. Instead, that information is stored in a central database on the Inari application server. Additionally, an object-oriented rules engine on the application server stores the rules for behavior of each node on the network. The application server, therefore, is a real-time repository of all objects and variables residing within the distributed nodes of an Inari system. Furthermore, the application server is responsible for inter-node communication by executing the rules of behavior between nodes as conﬁgured by the user of the system. Inari’s implementation of CAL allows the application server to perform tasks re- quiring greater intelligence, storage, and processing power. In so doing, the cost of each client node is greatly reduced. In other words, a light switch is allowed to be a light switch with a minimum amount of memory and logic. Consequently, a minimum cost of the system is reached because each network node only carries with it the minimal amount of functionality, circuitry, and intelligence it requires to perform its function whether that function is extremely simple or very complex. Finally, Inari CAL uses all of generic CAL’s deﬁned contexts, objects, variables, and methods. This allows the attributes of all Inari PNT devices to be exposed in a common format, facilitating interoperability and remote manageability.

Server Applications

Inari provides a critical set of server components that developers can leverage when creating home network servers or residential gateways for their customers. These server ingredients combine to provide the OEM with all the major software components, above the operating system, required to implement a residential gateway. These server technology components include the following: 114 6 Powerline-Based Home Networks

• Proxy server—The Inari proxy server routes and distributes Internet data, voice, and video, including VOIP, within the home, providing shared access to a single Internet connection. • E-mail server—Inari PNT offers OEMs an e-mail server solution to facilitate in-home and in-office mail, as well as providing automatic, unattend- ed delivery and retrieval of e-mail between the server and the customer’s ISP. • Network administration services—Inari’s network administration services enables customers to manage and secure their home or small office powerline network. • Network installation services—Creating a network on Windows-based computers is not a trivial task. Inari’s network installation services automatically loads the necessary protocol stacks, binds the protocol to the device, and configures the TCP/IP address on all of the networked computers. • Service provider interface—Inari’s service provider interface enables inter-device communication between the network nodes and the Inari services that reside on the server (i.e., proxy server, e-mail server, and network administration). Currently, Inari provides an Inari CAL interface for Win- dows, Win32, and LINUX. Inari is in the process of developing an interface for UPnP and plans to develop other interfaces that may include OSGi (Sun and Oracle) or emMicro (emWare).

SUMMARY ......

With the exception of battery-operated devices, everything electrical in your home is already connected to the powerline network. This makes the powerline a natural choice for home networking. Powerline networking allows you to connect your in- home appliances to the rest of the world through any electrical outlet. X-10 and CEBus are two popular technologies that deliver data over lines that previously delivered only electricity. A company called Enikia is providing a range of products and competing technologies that allow home electronics devices such as set-top boxes to connect to a powerline home network via the nearest electrical outlet. Inari PNT provides a set of network technology ingredients and components that enables OEMs that cater to the home/home ofﬁce network market to easily take advantage of the ubiquitous nature of the powerline infrastructure.